Optimization of Xylose Recovery in Oil Palm Empty Fruit Bunches for Xylitol Production

The hardest obstacle to make use of lignocellulosic biomass by using green technology is the existence of lignin. It can hinder enzyme reactions with cellulose or hemicellulose as a substrate. Oil palm empty fruit bunches (OPEFBs) consist of hemicellulose with xylan as the main component. Xylitol production via fermentation could use this xylan since it can be converted into xylose. Several pretreatment processes were explored to increase sugar recovery from lignocellulosic biomass. Considering that hemicellulose is more susceptible to heat than cellulose, the hydrothermal process was applied to OPEFB before it was hydrolyzed enzymatically. The purpose of this study was to investigate the effect of temperature, solid loading, and pretreatment time on the OPEFB hydrothermal process. The xylose concentration in OPEFB hydrolysate was analyzed using high-performance liquid chromatography (HPLC). The results indicated that temperature was more important than pretreatment time and solid loading for OPEFB sugar recovery. The optimum temperature, solid loading, and pretreatment time for maximum xylose recovery from pretreated OPEFB were 165 °C, 7%, and 60 min, respectively, giving a xylose recovery of 0.061 g/g of pretreated OPEFB (35% of OPEFB xylan was recovered).


Introduction
Xylitol can be found naturally in some fruits and vegetables in a small amount, and it is widely consumed by diabetic patients as a sugar substitute since it does not need insulin to be metabolized [1,2]. Xylitol also has low potential to produce tooth decay; thus, it is widely applied in toothpaste and confectionery. Xylitol can be biologically synthesized from xylan, the part of hemicellulose in lignocellulosic biomass. The lignocellulosic biomass should be hydrolyzed enzymatically to obtain xylose as the main substrate for xylitol fermentation.
Oil palm empty fruit bunches (OPEFBs) represent the main lignocellulosic biomass waste from the palm oil industry. For every ton of crude palm oil (CPO) produced, 0.96 tons of OPEFB is generated [3,4]. OPEFB is composed of cellulose 31%-43%, hemicellulose 23%-35%, and lignin 11%-23% [5][6][7][8], which are bound together by hydrogen and covalent bonds to form sturdy structures that are important for plants. Cellulose is one of the structural carbohydrates in OPEFB with glucan as the constituent, whereas xylan and arabinan are the constituents of hemicellulose [9]. temperature, and process time, and the three levels were coded as −1, 0, and +1. Table 1 shows the matrix for every variable level studied in this experiment. The experimental design matrix produced according to the Box-Behnken design consisted of 15 combinations and was carried out in duplicate, resulting in a total of 30 runs.

Hydrothermal Pretreatment
The pretreatments were carried out in a 100-mL autoclavable mini reactor (manufactured by a local workshop in Bandung, Indonesia) equipped with a paddle stirrer and stirrer cooler. The OPEFB used was mixed with water at various SL and heated at various temperatures and process times following the design experiment generated by Minitab ® 18.1. After the pretreatment process, the mixture was screened using a fine mesh cloth, and the pretreated solid was left in an oven at 50 • C for 24 h while the liquid was stored in the refrigerator before the hydrolysis process.

Enzymatic Hydrolysis
Hydrolysis of the pretreated solid was performed in a 100-mL Schott bottle at 5% solid loading in an incubator shaker (Daihan Labtech, Namyangju, South Korea). The citrate buffer solution, 0.05 M pH 5.0, was added until a working volume of 50 mL. Prior to the addition of enzyme, the mixture of OPEFB pretreated solid and buffer was autoclaved at 121 • C for 15 min. The liquid from the hydrothermal process was hydrolyzed in the same condition as the pretreated solid. The enzymatic hydrolysis was performed for 48 h at 50 • C and 150 rpm, using commercial xylanase (CAS No: 110-44-1) from Shandong Dianmei International Trade Co., Ltd. (Jinan, China) in a powder form with an activity of 22,072.48 U/g. Samples were centrifuged at 6000 rpm for 15 min and then filtered using a 0.22-µm syringe filter before xylose concentration analysis of the hydrolysate.

Sugar Analysis (Xylose)
Xylose concentration in the hydrolysate was analyzed using HPLC (Alliance HPLC System, Waters, Milford, MA, USA) using a method developed by National Renewable Energy Laboratory (NREL) [32]. The xylose was separated using an HPX-87H Biorad-Aminex column (length of 300 mm, a particle diameter of 9 µm, and an internal diameter of 7.8 mm), which was maintained at 60 • C and detected using a 2414 refractive index (RI) detector (detector temperature 40 • C). The column conditions were set as follows: sample volume 20 µL, a mobile phase of 0.005 M sulfuric acid, a flow rate of 0.6 mL/min, a detector temperature of 40 • C, and processing time of 30 min. Xylose as the standard in HPLC analysis was purchased from Sigma-Aldrich (Singapore), Chemicals for citrate buffer were purchased from E-Merck (Singapore).
Xylose recovery was calculated using Equation (

Statistical Analysis and Optimization
The optimization was performed using the response surface method and a Box-Behnken design. ANOVA analysis of the factors, i.e., a three-way ANOVA, was performed using MINITAB ® 18.1, in addition to designing experimental designs for optimization. In the optimization of the evaluated factors, the xylose yield in solid and liquid pretreated OPEFB was maximized as the objective function.

Results
The results are divided into the effects of temperature, solid loading, and pretreatment time on xylose recovery, and the optimization of temperature, SL, and pretreatment time to obtain maximum xylose recovery from pretreated OPEFB. A surface plot was generated to evaluate the effects of the three factors on xylose recovery using statistical software MINITAB ® 18.1.

The Effects of SL, Temperature, and Time on Xylose Recovery
The effects of SL, temperature, and time on xylose recovery in the hydrolysate of solid and liquid pretreated OPEFB are shown in Figures 1a,b, respectively.  The results in Figure 1a,b showed that SL, temperature, and pretreatment time affected the xylose recovery in solid and liquid pretreated OPEFB hydrolysate. Every single line showed a slope that represented effects of magnitude, where a steeper slope designates greater magnitude of the effects. The SL of 10% was associated with the maximum xylose recovery in the hydrolysate of solid pretreated OPEFB (Figure 1a), and the pretreatment time of 37.5 min was associated with the highest xylose recovery in solid pretreated OPEFB hydrolysate (Figure 1a). A temperature of 165 • C was associated with the highest xylose recovery as shown in Figure 1a,b. The SL of 10% and pretreatment time of 37.5 min showed the highest xylose recovery on the hydrolysate of solid pretreated OPEFB. On the contrary, the SL of 10% and pretreatment time of 37.5 min in Figure 1b showed the lowest xylose recovery in the hydrolysate of liquid pretreated OPEFB. To ensure the main effect in the plot, the ANOVA result is shown in Table 2. Analysis of variance (Table 2) showed that temperature had a significant effect with a p-value of 0.001 on the square model of the ANOVA test, which explained the curvature of xylose recovery and clear maximum recovery as responses in Figure 1a. However, the maximum xylose recovery in solid pretreated OPEFB hydrolysate associated with SL and pretreatment time may be due to random chance as the p-value was higher than 0.05. The linear model of ANOVA test showed that SL, pretreatment time, and temperature were not statistically significant for xylose recovery in solid pretreated OPEFB hydrolysate as the p-values were 0.786, 0.423, and 0.721, respectively. Likewise, the square model on xylose recovery showed that SL and pretreatment time were not as significant as temperature because the p-values were 0.258 and 0.293, respectively. Furthermore, there was no interaction effect between the effects and only a two-way interaction between temperature and time with a p-value of 0.314, which was not significant, as can be seen in Table 2.
The minimum xylose recovery in liquid pretreated OPEFB hydrolysate associated with SL and pretreatment time may be due to random chance ( Figure 1b). This was supported by the ANOVA test shown in Table 2. The analysis of variance showed that SL, temperature, and pretreatment time effects were not statistically significant as the p-values were 0.917, 0.041, and 0.817, respectively. There was no interaction effect between these three parameters on xylose recovery in solid and liquid pretreated OPEFB hydrolysate. There was only a two-way interaction between temperature and pre-time with a p-value of 0.377, which was not significant. However, the temperature had a significant effect on the square model, which confirmed the curvature in Figure 1a,b.

The Surface Plot of Xylose Recovery vs. Temperature and SL
This section describes the effect of temperature and SL at a pretreatment time of 37.5 min on xylose recovery as shown in Figure 2a,b. The surface plots for the effect of temperature and SL at 15 min and 60 min are not shown because they were similar. with a p-value of 0.314, which was not significant, as can be seen in Table 2.
The minimum xylose recovery in liquid pretreated OPEFB hydrolysate associated with SL and pretreatment time may be due to random chance (Figure 1b). This was supported by the ANOVA test shown in Table 2. The analysis of variance showed that SL, temperature, and pretreatment time effects were not statistically significant as the p-values were 0.917, 0.041, and 0.817, respectively. There was no interaction effect between these three parameters on xylose recovery in solid and liquid pretreated OPEFB hydrolysate. There was only a two-way interaction between temperature and pretime with a p-value of 0.377, which was not significant. However, the temperature had a significant effect on the square model, which confirmed the curvature in Figures 1a,b.

The Surface Plot of Xylose Recovery vs. Temperature and SL
This section describes the effect of temperature and SL at a pretreatment time of 37.5 min on xylose recovery as shown in Figures 2a,b. The surface plots for the effect of temperature and SL at 15 min and 60 min are not shown because they were similar. As described above, Figures 2a,b showed a curved shape in the response surface plot, which suggested that there was an optimum value. The surface plot justified the main effect plot in Figure  1. There was a slight increase in xylose recovery in solid pretreated OPEFB and a slight decrease in liquid pretreated OPEFB, along with an increase in SL up to 10%. As for temperature, the figure asserted that there would be a maximum temperature at about 165 ± 5 °C, corresponding to a As described above, Figure 2a,b showed a curved shape in the response surface plot, which suggested that there was an optimum value. The surface plot justified the main effect plot in Figure 1. There was a slight increase in xylose recovery in solid pretreated OPEFB and a slight decrease in liquid pretreated OPEFB, along with an increase in SL up to 10%. As for temperature, the figure asserted that there would be a maximum temperature at about 165 ± 5 • C, corresponding to a maximum xylose recovery. The higher SL in combination with temperature gave a higher xylose recovery and decreased after 165 • C. The peak on the plot corresponds to the highest xylose recovery in solid pretreated OPEFB hydrolysate at an approximate temperature of 165 ± 5 • C and SL of 15%. This combination of temperature and SL showed that there was a need for an optimization procedure.

The Surface Plot of Xylose Recovery vs. Pretreatment Time and SL
The effects of pretreatment time and SL at a constant temperature of 165 ± 5 • C on xylose recovery are shown in Figure 3. The surface plots for the effect of pretreatment time and SL at 130 ± 5 • C and 200 ± 5 • C are not shown because they were similar. The pretreatment time here means the time which proceeded while heating the mixture at a certain temperature. It should be noted that the reactor configuration used in this study did not allow it to be cooled immediately after heating, and it took about 22 h to cool down.

The Surface Plot of Xylose Recovery vs. Pretreatment Time and SL
The effects of pretreatment time and SL at a constant temperature of 165 ± 5 °C on xylose recovery are shown in Figure 3. The surface plots for the effect of pretreatment time and SL at 130 ± 5 °C and 200 ± 5 °C are not shown because they were similar. The pretreatment time here means the time which proceeded while heating the mixture at a certain temperature. It should be noted that the reactor configuration used in this study did not allow it to be cooled immediately after heating, and it took about 22 h to cool down.  Figures 3a,b show an opposite trend of xylose recovery; furthermore, there was a curvature of pretreatment time effect on xylose recovery in solid and liquid pretreated OPEFB hydrolysate. This curvature supposedly gave an optimum value of SL and pre-time. Using SL of 15% gave the highest xylose recovery in solid pretreated OPEFB hydrolysate. Furthermore, a combination of the longest pretreatment time of 60 min and SL of 15% gave the highest xylose recovery in solid pretreated OPEFB hydrolysate. On the contrary, xylose recovery in liquid pretreated OPEFB hydrolysate was lowered with the increase in SL and pretreatment time. The highest xylose recovery in liquid pretreated OPEFB hydrolysate was obtained at SL of 5% and a pretreatment time of 15 min. This contrast in pretreatment time and SL combination between solid and liquid pretreated OPEFB hydrolysate showed that optimization should be performed for each biomass matrix. The minimum peak of xylose recovery in liquid pretreated OPEFB hydrolysate shown in Figure 3b was obtained at the pretreatment time of 37.5 min and SL of 10%.

The Surface Plot of Xylose Recovery vs. Time and Temperature
The other combination effect on xylose recovery was pretreatment time and temperature at a constant SL of 10%, as presented in Figures 4a,b. The surface plots for the effect of time and temperature at SL of 5% and SL of 15% are not shown because they were similar.  Figure 3a,b show an opposite trend of xylose recovery; furthermore, there was a curvature of pretreatment time effect on xylose recovery in solid and liquid pretreated OPEFB hydrolysate. This curvature supposedly gave an optimum value of SL and pre-time. Using SL of 15% gave the highest xylose recovery in solid pretreated OPEFB hydrolysate. Furthermore, a combination of the longest pretreatment time of 60 min and SL of 15% gave the highest xylose recovery in solid pretreated OPEFB hydrolysate. On the contrary, xylose recovery in liquid pretreated OPEFB hydrolysate was lowered with the increase in SL and pretreatment time. The highest xylose recovery in liquid pretreated OPEFB hydrolysate was obtained at SL of 5% and a pretreatment time of 15 min. This contrast in pretreatment time and SL combination between solid and liquid pretreated OPEFB hydrolysate showed that optimization should be performed for each biomass matrix. The minimum peak of xylose recovery in liquid pretreated OPEFB hydrolysate shown in Figure 3b was obtained at the pretreatment time of 37.5 min and SL of 10%.

The Surface Plot of Xylose Recovery vs. Time and Temperature
The other combination effect on xylose recovery was pretreatment time and temperature at a constant SL of 10%, as presented in Figure 4a  As discussed in the previous section, Figures 4a,b show that there was curvature on the surface plot. The xylose recovery in solid and liquid pretreated OPEFB hydrolysate increased as pretreatment time increased, with a slight decrease for solid pretreated OPEFB after 37.5 min. The maximum xylose recovery in solid and liquid pretreated OPEFB hydrolysate was approximately at a temperature of 165 ± 5 °C and pretreatment time of 60 min. The results ensured that optimization should be performed in order to obtain an optimum combination of pretreatment time, temperature, and SL.

Optimization
The optimization study was conducted to evaluate the pretreatment process, temperature, pretreatment time, and SL of OPEFB for xylitol production which required high xylose recovery for economic reasons. As discussed in the previous section, Figure 4a,b show that there was curvature on the surface plot. The xylose recovery in solid and liquid pretreated OPEFB hydrolysate increased as pretreatment time increased, with a slight decrease for solid pretreated OPEFB after 37.5 min. The maximum xylose recovery in solid and liquid pretreated OPEFB hydrolysate was approximately at a temperature of 165 ± 5 • C and pretreatment time of 60 min. The results ensured that optimization should be performed in order to obtain an optimum combination of pretreatment time, temperature, and SL.

Optimization
The optimization study was conducted to evaluate the pretreatment process, temperature, pretreatment time, and SL of OPEFB for xylitol production which required high xylose recovery for economic reasons.
The results in previous sections suggested that optimization should be performed with a goal to maximize xylose recovery in solid and liquid pretreated OPEFB. This pretreated OPEFB would be used further in the hydrolysis process followed by fermentation for xylitol production. The dark green in the middle of the contour plot in Figure 5a,b shows the area of the highest total xylose recovery (>0.07 g/g of biomass). The range of optimal solid loading, temperature, and time values indicates that maximum xylose recovery ranged from 5%-15%, 155-178 • C, and 22-60 min. However, the optimization resulted in 43% xylose yielded from OPEFB xylan, which corresponds to 0.075 g/g biomass. The optimum process parameters were at a temperature of 165 ± 5 • C, with SL of 7% and a time of 60 min. This was validated at a temperature of 165 ± 5 • C, with SL of 7% and a pretreatment time of 60 min. The validation showed that 35% of xylose yielded from OPEFB xylan, which corresponds to 0.061 g/g of biomass. Even though the relative standard error of data was 8.4%, the validation yield of xylose from OPEFB xylan was 24% lower than that in the model. ANOVA results showed that the coefficients of determination (R 2 ) for xylose recovery in the pretreatment of solid OPEFB and liquid were 80.61% and 61.99%, respectively, which showed a deviation between the model and the results of the validation.

Discussion
Among the three variables studied, temperature gave the most prominent effect of xylose recovery both in the solid and the liquid pretreated OPEFB (Figures 1a,b). In agreement with the characteristics of hemicellulose, which is more susceptible to heat than other lignocellulosic components [33], an optimum value of xylose recovery was observed within the range of studied temperature. Hemicellulose tends to solubilize more than cellulose in a higher pretreatment temperature [24,26,34,35]. The soluble hemicellulose can be hydrolyzed further into furfural and become an inhibitor to enzymes [36]. The optimum temperature found in this study was in the same range of the optimum temperature obtained for the hydrothermal pretreatment of prairie cordgrass [24]. Other studies stated that the pretreatment of sugarcane bagasse, sugarcane straw, wood, and

Discussion
Among the three variables studied, temperature gave the most prominent effect of xylose recovery both in the solid and the liquid pretreated OPEFB (Figure 1a,b). In agreement with the characteristics of hemicellulose, which is more susceptible to heat than other lignocellulosic components [33], an optimum value of xylose recovery was observed within the range of studied temperature. Hemicellulose tends to solubilize more than cellulose in a higher pretreatment temperature [24,26,34,35]. The soluble hemicellulose can be hydrolyzed further into furfural and become an inhibitor to enzymes [36]. The optimum temperature found in this study was in the same range of the optimum temperature obtained for the hydrothermal pretreatment of prairie cordgrass [24]. Other studies stated that the pretreatment of sugarcane bagasse, sugarcane straw, wood, and OPEFB at temperatures higher than 165 • C resulted in dissolved hemicellulose [37][38][39][40]. However, our current study differs from previously reported results. The xylose recovery in liquid pretreated OPEFB was lower than in the solid pretreated OPEFB, as can be seen in Figures 2-4. During the hydrothermal process, some of the water was absorbed into the oil palm empty fruit bunches, leaving only a portion of the water, with a maximum of 50% of the initial volume. This led to the lower concentration of hemicellulose in the remaining liquid. A previous experiment using wheat straw in a continuous hydrothermal pretreatment process resulted in a hemicellulose extraction of 70% at 195 • C for 6-12 min [41]. Another experiment using catalpa wood resulted in a different result, which showed an extraction of 33.9% and 38.8% hemicellulose at 160 • C and 170 • C, respectively [42]. These results showed that there was a connection between lignocellulose type and hemicellulose on the extraction recovery and conditions, as reported earlier [43].
Solid loading (SL) effects on xylose recovery for pretreated solid and liquid were not statistically significant (p-values of 0.957 and 0.93, respectively) up to SL of 15%, as can be seen in Figure 1a,b. This is important because a higher SL (>15%) lowers the xylose recovery, as reported earlier [44,45]. Hydrothermal pretreatment uses water as the agent to break the bond between lignin and hemicellulose. A high SL means that there is less water in the system that should be in contact with lignocellulosic biomass, which may lower xylose recovery [46]. Another study stated that a high SL pretreatment (20% SL) required a combination pretreatment using maleic acid and a saccharification process [46]. Another difficulty in using high solid loading was obtaining a pumpable slurry for the subsequent enzymatic hydrolysis [47].
An increase in the pretreatment time resulted in a slight rise in xylose recovery in solid and liquid pretreated OPEFB hydrolysate, as can be seen in Figures 2 and 3. A further increase in xylose recovery was reported to be less important as time prolonged further. Extending the hydrothermal pretreatment time of OPEFB at 127.9 • C from 60 to 90 min was reported not to give a considerable increase in xylose recovery [23]. A similar report on OPEFB hydrothermal pretreatment showed that a longer pretreatment time resulted in a lower sugar recovery [44]. Furthermore, extending the hydrothermal pretreatment time to 180 min at 170 • C lowered the xylose recovery of beechwood [25]. A longer pretreatment time leads to the formation of furfural, soluble lignin, and insoluble lignin, which was reported to inhibit the enzymatic hydrolysis process [25,48]. Compared to previous research using grapevine followed by acid hydrolysis, where 60% of xylan was recovered in the liquid pretreated biomass [25]. Considering that melted lignin was not isolated before enzymatic hydrolysis was performed, the xylan recovery of 35% was adequate. A similar result was reported previously on hot water treatment using bamboo [35]. Other reports stated that enzyme can be absorbed by deposited lignin in the solid surface of biomass [49][50][51].
The low determination coefficient of xylose recovery for liquid pretreated OPEFB was due to the low xylose recovery, as can be seen in Figures 2b and 4b. Even though the model R 2 was 61.99%, it fit significantly (p-value of 0.001). The ANOVA results confirmed that there was no interaction effect among factors, SL, temperature, and pretreatment time. However, there was obviously a need for optimization for these three factors to obtain maximum xylose recovery in moderate conditions as can be seen in Figure 5. The optimum temperature was 165 ± 5 • C, consistent with the results obtained on hydrothermal pretreatment of prairie cordgrass [24]. The optimum SL of 7% was in good agreement with the previous result in hydrothermal pretreatment of OPEFB [44]. The optimum pretreatment time of 60 min was consistent with the results obtained on hot-compressed hot water pretreatment of sugarcane bagasse [52].

Conclusions
This research found that the temperature was the most important factor compared to solid loading and pretreatment time in the hydrothermal pretreatment process of OPEFB to obtain high xylan recovery. The maximum yield of xylose in solid and liquid pretreated OPEFB was obtained at a temperature of 165 • C, pretreatment time of 60 min, and SL of 7% with a yield of 0.06 g xylose/g biomass.